the federal reserve's balance sheet and earnings: a · pdf filevol. 11 no. 2 the federal...

The Federal Reserve’s Balance Sheet andEarnings: A Primer and Projections∗

Seth Carpenter, Jane Ihrig, Elizabeth Klee, Daniel Quinn,and Alexander BooteFederal Reserve Board

Over the past few years, the Federal Reserve’s use of uncon-ventional monetary policy tools has received a vast amount ofpublic attention, from discussing how these asset purchaseshave put downward pressure on longer-term interest rates andthus supported economic activity to evaluating the implica-tions for Federal Reserve remittances to the Treasury and theeffect on monetary and fiscal policy. As the economic recoveryhas gained some momentum of late, the focus has turned toissues associated with the normalization of monetary policy.In this paper, we begin by providing a primer for the Fed-eral Reserve’s balance sheet and income statement. With thatfoundation in place, we then consider a variety of scenariosconsistent with statements by Federal Reserve officials abouthow the FOMC will normalize policy, including whether tosell mortgage-backed securities, whether to change the com-position of Federal Reserve liabilities, and the timing of lift-ing the federal funds rate off from the zero lower bound. Ineach of these scenarios, we discuss the implications of thesenormalization policies on the size and composition of FederalReserve asset and liability holdings and on remittances of earn-ings to the Treasury, which capture the interest rate risk ofthese normalization policies. We show that under a baseline

∗The authors are staff economists and research assistants in the Division ofMonetary Affairs, Board of Governors of the Federal Reserve System, Washing-ton, D.C. 20551 U.S.A. We thank James Clouse, Bill English, Michelle Ezer, DonHammond, Lawrence Mize, Julie Remache, Viktors Stebunovs, Lisa Stowe, JeffMoore, Ari Morse, Brett Schulte, Joe Kachovec, and three anonymous refereesfor thoughtful comments and assistance. The views in this paper are solely theresponsibility of the authors and should not be interpreted as reflecting the viewsof the Board of Governors of the Federal Reserve System or of any other personassociated with the Federal Reserve System.

237

238 International Journal of Central Banking March 2015

normalization strategy described by policymakers, the balancesheet should slowly return to a more normal composition andsize, while remittances should remain sizable. With some alter-native normalization plans, especially if faced with high inter-est costs, remittances could drop to zero for some time.

JEL codes: E52, E58, E47.

1. Introduction

In response to the financial crisis that began in 2007 and the sub-sequent recession, the Federal Reserve employed a variety of non-traditional monetary policy tools that garnered a vast amount ofpublic discussion. Some discussion focused on the expanding sizeand changing composition of the Federal Reserve’s balance sheetand, specifically, the Federal Reserve’s holdings of securities in theSystem Open Market Account (SOMA) (Federal Reserve Bank ofNew York 2013). This expansion led to discussions about the effectsof unconventional monetary policy on interest rates (Krishnamurthyand Vissing-Jorgensen 2011, Li and Wei 2013). In addition, otherauthors highlighted implications for Federal Reserve transfers to theTreasury (“remittances”), the effect on monetary and fiscal policy(Rudebusch 2011, Greenlaw et al. 2013), and the independence ofthe Federal Reserve from political pressure (Christensen, Lopez, andRudebusch 2013).

More recently, as the economic recovery has gained some momen-tum, the discussion has turned to questions about the normalizationof monetary policy. In various venues, Federal Open Market Com-mittee (FOMC) participants have expressed their views about nor-malizing the stance of monetary policy. In particular, the June 2013FOMC minutes provided some discussion on policy normalizationand the long-run composition of the balance sheet, while FOMCstatements clearly tie the rise in the federal funds rate to the outlookfor unemployment and inflation. In this paper, we consider how theFederal Reserve’s balance sheet, and the income that derives fromthe balance sheet, might evolve under a variety of assumptions aboutthe path of monetary policy and approaches to the normalization ofpolicy. For example, we consider the June 2011 exit principles thatincluded sales of mortgage-backed securities (MBS) as part of the

Vol. 11 No. 2 The Federal Reserve’s Balance Sheet and Earnings 239

normalization process, as well as the more recent information laidout in the June 2013 minutes that suggests that such sales would notbe a prominent part of the early stages of policy normalization. Inaddition, given the evolving views in markets about the likely timingof the first increase in the federal funds rate, we consider a scenariowhere the date of liftoff is pushed out, consistent with some variantof quantitative forward guidance related to an unemployment rate orinflation threshold, and analyze the effect of that timing for the pathof the balance sheet. Finally, we discuss some of the possible implica-tions for Federal Reserve expenses from choosing different mixes ofFederal Reserve liabilities (known as “reserve-draining tools”) duringthe normalization process.

The scenarios presented here do not provide an exhaustive rangeof options that the FOMC could use to remove monetary policyaccommodation. They do, however, show how many tools, in isola-tion, can aid the Committee in reducing the size of the balance sheetand boost short-term interest rates.1 With these scenarios in mind,one can then see the impact of any convex combination of tools thatthe Committee may choose during the normalization process.

In analyzing each normalization scenario, we report the lengthof time until the Federal Reserve’s balance sheet returns to a nor-mal size. We also project how MBS holdings will evolve, given thatholdings of MBS are a particularly novel development for the Fed-eral Reserve and minutes from FOMC meetings suggest that theiracquisition has been a source of some debate. In addition, we lookat the interest rate risk of different exit strategies that appear tobe under consideration. Such considerations may be important if, asGreenlaw et al. (2013) suggest, a period of zero remittances resultsin negative political pressures.

The remainder of the paper is organized as follows. Section 2 pro-vides a primer on the Federal Reserve’s balance sheet and incomestatement. Section 3 outlines the scenario assumptions used asinputs to the projections of the balance sheet. The baseline balancesheet and income projections are discussed in section 4. Section 5considers the alternative normalization policies. Section 6 providesthe sensitivity analysis. Section 7 concludes.

1See Ihrig et al. (2012) for how to link the size of the balance sheet to monetarypolicy accommodation.


2. The Federal Reserve’s Balance Sheet, IncomeStatement, and Valuation of the SOMA Portfolio

In this section, we review key balance sheet components in our pro-jections, as well as the income generated from the balance sheet. Wealso provide some historical context for the evolution of these items.

2.1 The Federal Reserve’s Balance Sheet

Our discussion of the Federal Reserve’s balance sheet will refer to theconsolidated balance sheets of the twelve individual Reserve Bankbalance sheets.2 In reality, the accounting that will be discussedbelow is done at the Reserve Bank level; however, for simplicity, wefocus on the Federal Reserve System’s aggregate balance sheet.

Like any balance sheet, the Federal Reserve has assets on oneside of the balance sheet, which must equal liabilities plus capital onthe other side. As shown in table 1, at the end of 2006, total assetsof the Federal Reserve were $874 billion, with the single largest assetitem being the SOMA portfolio, at about $780 billion. Prior to thefinancial crisis, the domestic SOMA portfolio comprised only Treas-ury securities, of which roughly one-third were Treasury bills andtwo-thirds were Treasury coupon securities. On the other side ofthe balance sheet, the largest liability item was paper currency, orFederal Reserve notes (FR notes), at about $785 billion.

With the lending that took place during the financial crisis, fora time, the amount outstanding in the credit and liquidity facili-ties surpassed the size of the SOMA portfolio. As of December 25,2013, however, the SOMA portfolio was again the largest asset item,and it had grown to $3.8 trillion because of the asset purchase pro-grams. On the liability side of the balance sheet, FR notes, at about$1.2 trillion, were no longer the largest liability item. Instead, asthe FOMC increased its asset purchases, reserve balances increasedcorrespondingly to a level of about $2.5 trillion.

2The Board of Governors does not hold assets and liabilities in the same waythat the Reserve Banks do. Section 10 of the Federal Reserve Act authorizesthe Board to levy semi-annually upon the Reserve Banks, in proportion to theircapital stock and surplus, an assessment sufficient to pay its estimated expensesfor the half of the year succeeding the levying of such assessment, together withany deficit carried forward from the preceding half-year.


Table 1. Federal Reserve’s Balance Sheet

Balance Sheet End-2006 Balance Sheet End-2013

$Billion $Billion

Assets Liabilities Assets Liabilities

SOMA 779 Deposits ofDep. Inst.

13 SOMA 3,763 Deposits ofDep. Inst.

2,451

OtherAssets

95 FederalReserveNotes

783 OtherAssets

270 FederalReserveNotes

1,195

OtherLiabilities

48 OtherLiabilities

332

TotalCapital

31 TotalCapital

55

Source: H.4.1. statistical release.

The next few sub-sections review the key components of theFederal Reserve’s balance sheet and how they have changed.3

2.1.1 The SOMA Portfolio: Composition, Size, and MaturityStructure

Over most of the post-war period, the SOMA portfolio was thelargest asset item on the Federal Reserve’s balance sheet.4 Duringthat time, the SOMA portfolio essentially held Treasury securities;however, the portfolio has held other types of securities over thecourse of its history.5 For example, from 1971 to 1981, the FederalReserve purchased limited quantities of agency securities; the last ofthese securities matured in the early 2000s, and none was purchaseduntil 2008.6

3For a description of additional components of the bal-ance sheet, see the interactive guides to the H.4.1 tables athttp://www.federalreserve.gov/monetarypolicy/bst fedsbalancesheet.htmor the Financial Accounting Manual for Reserve Banks athttp://www.federalreserve.gov/monetarypolicy/files/bstfinaccountingmanual.pdf.

4For a description of the Federal Reserve’s balance sheet prior toWorld War II, see Banking and Monetary Statistics, 1914–1941, athttp://fraser.stlouisfed.org/title/?id=38.

5Refer to Edwards (1997).6Refer to Meltzer (2010).

http://www.federalreserve.gov/monetarypolicy/bst_fedsbalancesheet.htm

http://www.federalreserve.gov/monetarypolicy/files/bstfinaccountingmanual.pdf.

http://fraser.stlouisfed.org/title/?id=38.


Figure 1. SOMA, Capital + FR Notes, andReserve Balances

1990 1993 1996 1999 2002 2005 2008 2011

0

1000

2000

3000

4000$Billion

SOMACapital+NotesReserve Balances

Source: H.4.1 statistical release.

Historically, the size of the SOMA portfolio—and the balancesheet, more generally—reflected growth in FR notes and ReserveBank capital. When currency is put into circulation, it is shipped toa depository institution, and that institution’s account at the Fed-eral Reserve is debited by an equivalent amount. Because currencyoutstanding tends to trend upward, over time currency growth wouldtend to reduce the amount of reserve balances in the banking sys-tem. The Federal Reserve would purchase securities in open-marketoperations to offset this drain of reserves. On net, therefore, thegrowth rate of currency tended to drive the size of the balance sheet.Similarly, when a depository institution is required to subscribe toa larger amount of Federal Reserve capital or the Federal Reserveadds to its surplus account, the result would be, all else equal, areduction in reserve balances.7 As a result, the SOMA portfolio mustincrease to offset these increases as well, creating a larger balancesheet overall.

This historical pattern is illustrated in figure 1. As can be seen,through 2007, both the SOMA portfolio and currency and capital

7As will be more fully explained later in the paper, each member bank of aReserve Bank is required to subscribe to the capital of its district Reserve Bankin an amount equal to 6 percent of its own capital stock.


Figure 2. Weighted Average Maturity of SOMA

1980 1984 1988 1992 1996 2000 2004 2008 2012 0

20

40

60

80

100

120

140Months

Source: Federal Reserve Bank of New York.Notes: Includes only nominal Treasury securities.

trended upward together. When the asset programs began in late2008 and early 2009, and continuing through the second round ofpurchases in 2010 and 2011, the SOMA portfolio increased markedlyand at a rate that far outpaced the growth of currency and capital.With the initiation of the maturity extension program in 2011, thesize of the portfolio remained roughly constant; however, as depictedin figure 2, the weighted average maturity of Treasury securitiesin the SOMA portfolio increased markedly. From a longer perspec-tive, over time, the SOMA portfolio has had a range of maturitiesof Treasury securities in its holdings.8 Prior to the financial cri-sis, the Open Market Trading Desk (the Desk) tended to purchasesecurities across the entire yield curve to avoid distorting the yieldcurve. However, after the start of the financial crisis, the maturity ofTreasury coupon securities in the SOMA portfolio lengthenednotably, reflecting the runoff in bills to sterilize the credit and liquid-ity programs in 2008, and the purchase of longer-dated securitiesmore recently.

8In the weekly H.4.1 statistical release, in addition to the Federal Reserve’sbalance sheet, the maturity distribution of asset holdings is also published.


2.1.2 Deposits of Depository Institutions

Deposits of depository institutions include all depository institu-tions’ balances at the Federal Reserve that are used to satisfy reserverequirements and balances held in excess of balance requirements.Deposits of depository institutions grew dramatically through thecrisis, and are currently quite elevated by historical standards. Whenwe refer to “reserve balances,” we are using the “deposits of depos-itory institutions” concept. These deposits represent funds thatdepository institutions own; they are a liability of the Reserve Bank,but an asset of the depository institution. These funds are also usedfor payment system settlement; for example, a payment from onebank to another (or from one bank’s customer to the customer ofa different bank) typically results in a debit to the paying bank’saccount and a credit to the receiving bank’s account. Lending ofreserve balances and payment activity result only in a movementof reserve balances from one depository institution’s account atthe Federal Reserve to another institution’s account; the aggregatequantity is unchanged.

2.1.3 Federal Reserve Notes

Federal Reserve notes, or currency, are a liability of the FederalReserve. As a practical matter, the Federal Reserve does not deter-mine the quantity of currency outstanding. Instead, when a depos-itory institution wants to hold currency in its vault or automaticteller machines in order to meet customer needs, it requests a ship-ment from its Federal Reserve Bank. When that shipment is made,the depository institution’s reserve account at the Reserve Bankis debited by the amount of the currency shipment. One impor-tant source of demand for U.S. currency is from overseas. Althoughit is impossible to know with certainty what portion of currencyoutstanding is outside of the United States, estimates suggest thatthe fraction is one-half or more.9 Prior to the financial crisis, cur-rency was the largest liability item on the Federal Reserve’s balancesheet.

9Refer to Judson and Porter (1996).


2.1.4 Capital Paid In, Surplus, and Interest on FederalReserve Notes Due to U.S. Treasury

The capital of the Reserve Banks is different from the capital of otherinstitutions.10 It does not represent controlling ownership as it wouldfor a private-sector firm. Ownership of the stock is required by law,the Reserve Banks are not operated for profit, and the stock may notbe sold, traded, or pledged as security for a loan. As stipulated insection 5 of the Federal Reserve Act, each member bank of a ReserveBank is required to subscribe to the capital of its district ReserveBank in an amount equal to 6 percent of its own capital stock.Of this amount, half must be paid to the Federal Reserve Banks(referred to as capital paid in) and half remains subject to call bythe Board of Governors. This capital paid in is a required assessmenton the member banks and its size changes directly with the capitalof the member banks. Also stipulated by law is that dividends arepaid at a rate of 6 percent per year. Over the past decade, reflect-ing increases in capital at member banks, Reserve Bank capital hasgrown at an average rate of almost 15 percent per year. In addition,Reserve Banks have surplus capital, which reflects withheld earn-ings, and Federal Reserve Bank accounting policies stipulate thatthe Reserve Banks withhold earnings sufficient to equate surpluscapital to capital paid in. As a result, as capital of member banksgrows through time, capital paid in grows in proportion. Becausesurplus is set equal to capital paid in, it likewise grows at the samerate as member bank capital.

2.1.5 Deferred Asset

One liability item is distinct from the others. Under the FederalReserve’s remittance policy, the Federal Reserve remits all netincome to the U.S. Treasury, after expenses and dividends and allow-ing for surplus to be equated to capital paid in. As those earningsaccrue, they are recorded on the Federal Reserve’s balance sheet as

10See the Financial Accounting Manual for FederalReserve Banks, which reports the accounting standardsthat should be followed by the Federal Reserve Banks, athttp://www.federalreserve.gov/monetarypolicy/files/bstfinaccountingmanual.pdf,pages 1–68.

http://www.federalreserve.gov/monetarypolicy/files/bstfinaccountingmanual.pdf


“Interest on Federal Reserve notes due to U.S. Treasury.” In theevent that earnings only equal the amount necessary to cover oper-ating costs, pay dividends, and equate surplus to capital paid in, thisliability item would fall to zero because there are no earnings to remitand payments to the Treasury would be suspended. If earnings areinsufficient to cover these costs—that is, there is an operating lossin some period—then no remittance is made until earnings, throughtime, have been sufficient to cover that loss. The value of the earn-ings that need to be retained to cover this loss is called a “deferredasset” and is booked as a negative liability on the Federal Reserve’sbalance sheet under the line item “Interest on Federal Reserve notesdue to U.S. Treasury.” A deferred asset is an asset in the sense thatit reflects a reduction of future liabilities to the U.S. Treasury.

One consequence of the current implementation of FederalReserve Bank accounting policy is that the recording of a deferredasset implies that Reserve Bank capital does not decline in the eventof an operating loss. From time to time, individual Reserve Bankshave reported a deferred asset; for example, as shown on the H.4.1statistical release from November 3, 2011, the Federal Reserve Bankof New York recorded a deferred asset that week and the subsequentweek.11 However, by the third week, remittances had cumulated toa sufficient level to be able to pay off the deferred asset. More gener-ally, it has never been the case that the Federal Reserve System as awhole has suspended remittances to the Treasury for a meaningfulperiod of time because of operating losses.

Because there has never been a deferred asset of any signifi-cant size, there is little guidance as to the whether or not thereis a limit to the potential size of the asset. It may be plausibleto assume that it would not be allowed to exceed the value of allfuture earnings, possibly in present discounted terms, given the factthat it is paid down through future earnings. As will be clear inthe following projections, a scenario that would result in a deferred

11In November 2011, the Maiden Lane accounts, which are marked to marketand consolidated onto the balance sheet of the Federal Reserve Bank of NewYork (FRBNY), were revalued and resulted in an unrealized loss that requiredthe Federal Reserve Bank of New York to record a deferred asset. Over time, theFRBNY’s loans to the Maiden Lane limited liability companies were repaid infull, with interest.


asset in excess of the present value of future earnings is difficult tocontemplate.

Some foreign central banks do not record deferred assets andinstead use different accounting policies. For example, many foreigncentral banks smooth remittances each year, by transferring an aver-age amount of net income back to the government and saving the“excess” net income for times with negative shocks. Other foreigncentral banks allow for negative remittances—that is, transfers fromrather than to the government—if the loss is too large. The infusionof funds from their governments in cases of large negative shocksavoids deferred assets for these institutions. One example of a cen-tral bank with a form of a deferred asset is the Czech National Bank.This institution has operated for a number of years with a negativeequity position and zero remittances. We will return to this balancesheet line item in the projections, as we see what policy levers mayinduce a deferred asset.

2.2 The Federal Reserve’s Income Statement

As the Federal Reserve’s balance sheet has expanded in recent years,the income derived from the balance sheet has also grown, thoughthe key line items from the balance sheet that generated this incomeare the same. As shown in table 2, net income in both 2006 and 2013was driven by interest income from the SOMA portfolio. Despitethe difference in magnitude, in both years, SOMA interest incomewas more than 95 percent of total income. That said, SOMA inter-est income grew substantially over this period as the SOMA port-folio expanded. Interest expense, on the other hand, was minimalin both years. In particular, FR notes are a large liability withoutan associated interest expense. And, although the Federal Reservehas paid interest on reserve balances since October 2008, this lia-bility item has incurred little interest expense because the IOER(interest on excess reserves) rate has been at 25 basis points sinceDecember 2008. In both years, other items in the income statementwere similar. In total, remittances to the Treasury were positive inboth years, but much larger in 2013 because of the expanded SOMAportfolio.

The next few sub-sections review the key line items of the FederalReserve’s income statement in more detail.


Table 2. Federal Reserve’s Income and Expense

Income and Expense, 2006 Income and Expense, 2013$Billion $Billion

Income Expense Income Expense

Interest 37 Interest Expense 1 Interest 90 Interest Expense 5Income Income

Other 2 Other Expense 4 Other 1 Other Expense 7Income Income

Additions/ 4 Additions/ 2Deductions, Deductions,Dividends, and Dividends, andTransfers Transfers

Source: Federal Reserve annual report and press release titled “Reserve Bank Incomeand Expense Data and Transfers to the Treasury for 2013” (January 10, 2014).

2.2.1 SOMA Interest Income

As noted above, income on the securities held in the SOMA portfo-lio constitutes the vast majority of interest income. SOMA interestincome primarily reflects the size of the portfolio and the weightedaverage coupon (WAC) of the portfolio, less any amortized net pre-miums paid on securities.12 Prior to the financial crisis, the size ofthe portfolio increased steadily at a moderate rate. With the adop-tion of the asset programs, the securities portfolio expanded rapidlyand now stands at a level noticeably above its longer-run trend.The WAC, as shown in figure 3, fluctuated over time, rising andfalling with the market rates and the SOMA portfolio’s holdings.This pattern primarily reflects the fact that the Federal Reservereinvests maturing Treasury securities at auction, and the couponat auction tends to be in line with market rates. Although the assetpurchase programs resulted in a significant accumulation of longer-term debt in recent years, much of it was issued in a low interest

12SOMA interest income is defined as the rate of return on the portfolio (theproduct of the size of the portfolio times the WAC) minus amortized net premi-ums. Net premiums, though important in deriving the precise value of interestincome, will not be a primary driver of the contour of the projections of interestincome.


Figure 3. Weighted Average Coupon of SOMA

1980 1984 1988 1992 1996 2000 2004 2008 2012 0

2

4

6

8

10Percent

Source: Federal Reserve Bank of New York.Notes: Includes only nominal Treasury securities.

rate environment and, therefore, the WAC of the portfolio decreasedsomewhat.

Putting the size of the portfolio and the WAC of the portfo-lio together, as shown in figure 4, interest income climbed at amoderate pace in the years prior to the financial crisis, primarilybecause of the steady increase in the size of SOMA, which rose inline with the growth of FR notes and capital. Beginning in 2009,interest income from the portfolio rose noticeably as large-scale assetpurchases increased the size of the portfolio.

2.2.2 Interest Expense

With the introduction of interest on reserves in the fall of 2008 andthe concurrent rise in the level of reserve balances, interest expenserose. As mentioned above, the IOER rate has been 25 basis pointssince December 2008, and as a result, even with a substantial vol-ume of reserve balances, interest expense from reserve balances hasbeen low compared with interest income and was roughly $5 billionin 2013.

In addition to interest expense from reserve balances, there is alsointerest expense from reverse repurchase agreements (RRPs), mostly


Figure 4. Interest Income

1980 1985 1990 1995 2000 2005 2010 0

20

40

60

80

100$Billion

Total Interest IncomeSOMA Interest Income

Source: Annual report of the Federal Reserve Board of Governors.

generated by the foreign repurchase agreement (RP) pool.13,14 Inter-est rates paid on the foreign RP pool are generally in line withmarket rates, and when reserve balances are relatively low, interestexpense on the foreign RP pool can represent a large share of totalinterest expense.

Reverse repurchase agreements with primary dealers and otherinstitutions and the Term Deposit Facility (TDF) also have associ-ated interest expense. In addition to the primary dealers, the FederalReserve selected money-market mutual funds, the Federal HomeLoan Mortgage Corporation (Freddie Mac), the Federal NationalMortgage Association (Fannie Mae), and some banks as potentialcounterparties for RRPs (both overnight and term). In contrast

13Before December 13, 2002, repo transactions were conducted as matchedsale-purchase transactions, where the Federal Reserve sold a security with anagreement to purchase it again at a later date. However, because matched sale-purchase transactions were accounted for as an outright sale rather than as afinancing transaction the way reverse repurchase agreements are, the transactionsdid not result in interest expense.

14Every business day, the Federal Reserve conducts overnight reverse repos withforeign central banks that hold dollars in their accounts at the Federal ReserveBank of New York. These transactions are one of the services that central banksprovide one another to facilitate their international operations.


to the RRPs, only banks are counterparties in TDF transactions.Although the Federal Reserve has developed the capability of con-ducting large-scale operations in either RRPs or the TDF, theseoperations have been only in the testing phase to date, and as aresult, interest expense associated with these operations has beenminimal.

2.2.3 Capital Gain (Loss)

Under Federal Reserve accounting rules, a Federal Reserve Bankrealizes gains or losses on a security only when the security is sold.At sale, the Federal Reserve’s gain or loss is the market value minusthe par value and unamortized net premiums on the security. Histor-ically, the Federal Reserve did not generally sell securities, becausethe secular growth in currency resulted in a need for a long-termincrease in securities holdings. In 2008, however, the Desk did sellsome securities to offset the expansion of the balance sheet thatresulted from the introduction of the liquidity facilities at the earlystages of the financial crisis. In that year, the Federal Reserve real-ized a capital gain of roughly $3 billion because market rates hadfallen, pushing up the market price of the securities sold. With thematurity extension program, the Federal Reserve also sold securities;in 2011, these sales realized a $2.3 billion capital gain.

2.2.4 Payment of Dividends, Transfers to Surplus, andInterest on Federal Reserve Notes Due to U.S. Treasury

As noted above, member banks are required to subscribe to thecapital stock of the Reserve Banks, and the Federal Reserve Actstipulates that the Federal Reserve pay a 6 percent dividend on thiscapital. Under policy prescribed by the Board of Governors, excessearnings are retained as surplus capital in an amount equal to capi-tal paid in. Before remittances to the Treasury are made, dividendsare paid and earnings are retained to equate surplus to capital paidin. Dividends are paid even if remittances to the Treasury would bezero. As discussed earlier, in the event that earnings fall short ofthe amount necessary to cover operating costs, pay dividends, andequate surplus to capital paid in, the Federal Reserve books a liabil-ity of “Interest on Federal Reserve notes due to U.S. Treasury.” Thisline item is recorded in lieu of reducing the Reserve Bank’s surplus.


Figure 5. Federal Reserve Distributions tothe U.S. Treasury

0

10

20

30

40

50

60

70

80

90$Billion

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013*

Source: Annual report of the Federal Reserve Board of Governors.*Preliminary unaudited estimate; see http://www.federalreserve.gov/newsevents/press/other/20140110a.htm.

2.2.5 Remittances to the Treasury

Each week, the Federal Reserve remits any earnings in excess ofoperating expenses and dividends to the U.S. Treasury.15 The use ofthese funds is stipulated in the Federal Reserve Act, which states:

The net earnings derived by the United States from FederalReserve banks shall, in the discretion of the Secretary, be usedto supplement the gold reserve held against outstanding UnitedStates notes, or shall be applied to the reduction of the out-standing bonded indebtedness of the United States under reg-ulations to be prescribed by the Secretary of the Treasury.16

Over time, as shown in figure 5, annual remittances remainedin a relatively small range, averaging about $25 billion in the years

15Occasionally, statutory transfers occur, which mandate that the FederalReserve transfer a portion of its surplus to the Treasury. The last time thisoccurred was in 2000, when approximately $3.8 billion held in the surplus accountwas transferred to the Treasury.

16Federal Reserve Act, Section 7, Use of Earnings Transferred to the Treasury,12 USC 290, sub-section (b).


Figure 6. Selected Treasury Receipts

1996 1999 2002 2005 2008 2011

0

500

1000

1500

2000$Billion

Fed EarningsSocial SecurityIndividual Income TaxesCorporate Income Taxes

Source: U.S. Treasury Bulletin.

immediately preceding the financial crisis. During the crisis, as Fed-eral Reserve income increased notably, so did remittances to theTreasury. Still, remittances remained a relatively small share ofgovernment receipts—dwarfed by individual income and corporateincome taxes, as shown in figure 6, and about in line with customsdeposits (not shown).

3. Projections Assumptions

In order to construct projections of the Federal Reserve’s balancesheet, assumptions about many of the details of the macroecon-omy as well as the Federal Reserve’s balance sheet and its evolutionmust be made. In addition, the projections in this paper are con-structed to be consistent with the Federal Reserve accounting prin-ciples discussed in section 2.17 The following sub-sections review keyassumptions made to project the balance sheet and income.

17The Federal Reserve’s accounting principles are published onthe website of the Board of Governors of the Federal Reserve:http://www.federalreserve.gov/monetarypolicy/files/bstfinaccountmanual.pdf.

http://www.federalreserve.gov/monetarypolicy/files/bstfinaccountmanual.pdf.


3.1 Scenario Assumptions and Results Overview

Our projections rely on the FOMC’s guidance regarding monetarypolicy normalization principles, the forecasts in the February 2014primary dealer survey conducted by the Federal Reserve Bank ofNew York, and the December 2013 and February 2014 Blue Chipforecasts. In the near term, we assume large-scale asset purchasesthat are in line with the median projection from the dealer sur-vey, with purchases in 2013 and 2014 totaling about $1.5 trillion.Consistent with the June 2011 FOMC “exit principles,” which weredetailed in the June 2011 FOMC meeting minutes, we assume thatthe first step to normalize the stance of monetary policy involvesthe FOMC allowing SOMA holdings to mature or prepay withoutreinvestment. Beyond that first move, we analyze a variety of alter-native normalization policies mentioned above. A summary of thekey results is shown in table 3.

In the baseline projection, we assume no MBS sales. The size ofthe SOMA portfolio will normalize by June 2021. Despite the nor-malization of the size of the portfolio, the composition of the port-folio will still reflect the non-traditional policy choices; at the end ofour projection period in 2025, over $500 billion of MBS will remainon the Federal Reserve’s books. The amount of these MBS holdingseven at this late date is still large, and residual sales would mostlikely take some careful consideration if sales were desired. Annualremittances to the Treasury are projected to remain sizable over thenear term and cumulate from 2009 through 2005 to about $920 bil-lion. Overall, this scenario suggests that large-scale asset purchaseswill have a net positive effect on income relative to a scenario withno purchases, but the Federal Reserve will continue to hold sizableMBS for some time.

The second scenario considers MBS sales. Under the June 2011exit strategy principles, sales of MBS were included because of adesire to return to a Treasury-only portfolio.18 Sales of MBS overfour years accelerate the date of normalizing the size of the portfo-lio by about two years relative to the scenario with no MBS sales.

18In the minutes of the April 2011 FOMC meeting, the reason for sellingMBS was to “minimize the extent to which the SOMA portfolio might affectthe allocation of credit across sectors of the economy.”


Tab

le3.

Sum

mar

yof

Alter

nat

ive

Nor

mal

izat

ion

Pol

icie

s

SO

MA

2009

–202

5Tro

ugh

SO

MA

Siz

eC

ompos

itio

n20

25M

BS

Cum

ula

tive

Rem

itta

nce

sN

orm

aliz

esN

orm

aliz

esH

oldin

gsR

emit

tance

s(D

ate)

Dat

e$B

illion

Bas

elin

eJu

n.20

21A

ug.20

22$5

38$9

18$1

8(2

018)

MB

SSa

les

May

2019

Aug

.20

20$0

$820

$0(2

017–

2020

)Lat

erLift

offN

ov.20

21N

ov.20

22$6

12$9

89$2

4(2

019)

Res

erve

-Dra

inin

gJu

n.20

21A

ug.20

22$5

38$8

75$1

1Too

ls+

50bp

(201

8)H

ighe

rIn

tere

stR

ates

——

——

—+

200b

pB

asel

ine

Sep.

2021

Aug

.20

22$6

15$7

91$0

(201

7–20

20)

MB

SSa

les

Aug

.20

19Ja

n.20

21$0

$634

$0(2

016–

2023

)


However, sales of MBS would also likely result in realized capitallosses on the MBS, an outcome that would most likely reduce annualremittances to zero for a few years. In pursuing this normalizationstrategy, the FOMC presumably would need to evaluate, amongother considerations, the trade-off of quickly reducing MBS hold-ings to zero with the possibility that remittances could be halted.In addition, seasoned MBS may have a coupon that is very differentfrom prevailing market interest rates, suggesting that these MBSwould need to be sold in a less liquid market than the one in whichthey were purchased, which might also be seen as risking unnec-essary volatility in these fixed-income markets at the critical pointwhen the FOMC is trying to firm the stance of policy.19

The third scenario explores a policy option that was discussed inearly 2014. As the unemployment rate began to decline toward 6.5percent, the Committee adjusted the quantitative thresholds thatdictated the forward guidance on the liftoff of the federal funds rateto language that would signal an extended period of time before rais-ing short-term interest rates.20 If this guidance implied a later liftoffthan what is in the baseline scenario, this would delay the date ofnormalization of the size of the balance sheet somewhat. Moreover,this alternative path for the balance sheet combined with a differentpath for interest rates would have implications for Federal Reserveincome and, as a result, remittances to the Treasury. In our analysis,a delay in liftoff would boost remittances but result in more MBSholdings at the end of the projection period, implying holding thesenon-traditional assets on the balance sheet for longer, or having tosell more residual MBS at some date in the future.

Finally, we examine the use of term reserve-draining tools. Thebaseline analysis does not explicitly model reserve-draining tools.One interpretation of this assumption is that no such tools are

19“Seasoned” MBS, or MBS that have been issued sometime prior, would needto be sold in the specified pool MBS market. As discussed in Vickrey and Wright(2013), and detailed in Friewald, Jankowitsch, and Subrahmanyam (2014), thetransaction costs of selling these securities far exceed those in the TBA mar-ket, where the securities were purchased. Executing a high volume of trades inany market with high transaction costs could potentially present difficulties formarket functioning.

20See the January and March 2014 FOMC minutes for a summary of theCommittee’s discussion.


needed, or that the use of RRPs or term deposits by the FederalReserve would be at the same cost as IOER.21 FOMC communi-cations suggest that policymakers are considering the use of thesereserve-draining tools during normalization. And, it is possible thatsome of the operations will involve transactions with terms longerthan overnight, which would most likely be at a rate that is abovethe federal funds rate. A priori, we have little information to gaugethe likely cost of these tools. The cost will depend on the rate as wellas the quantity of reserves that shift to these alternative tools. Toprovide a rough gauge as to how costly these tools could be, we assessthe effects on Federal Reserve net income if the interest expense is 50basis points higher than the projected level of the federal funds rateand apply this to all reserve balances. Fifty basis points is roughlyone standard deviation of the historical spread between the federalfunds rate and the yield on the three-month Treasury bill; assumingall reserve balances are drained to the higher expense tools providesan upper bound on this expense. Although interest expenses rise,there is only a modest effect on the Federal Reserve’s cumulativeremittances.

The analysis assumes that interest rates follow the median pathsforecasted by the primary dealers and Blue Chip respondents. Toexplore the interest rate sensitivity of our results, we also consider acase where interest rates are 200 basis points higher after liftoff, forboth the baseline and MBS sales scenarios. These results provide arough notion of the interest rate risk embedded in the SOMA portfo-lio. Compared with the baseline, the higher interest rate path impliesgreater interest expense on reserve balances, lower net income, andconsequently lower remittances to the Treasury. With no MBS sales,we find remittances to the Treasury are halted for three years. Inthe case of MBS sales, remittances are halted for eight years. Ofcourse, if sales were being implemented and rates rose dramaticallyas suggested here, policymakers could slow the pace of sales or stopthem entirely. In fact, if rates rose, sales, which would put upwardpressure on rates, might not be the preferred policy. In addition, theFOMC has said sales are not part of the normalization plan. Finally,

21The Federal Reserve has been testing overnight RRP operations since late2013. Since this tool is overnight, it most likely has a similar expense to IOERand is not discussed here.


in thinking about this scenario, it is important to put the shock insome perspective. Christensen, Lopez, and Rudebusch (2013) reportthat this interest rate shock scenario is very unlikely.

The sections that follow explore these conclusions in more detail.

3.2 Interest Rate Assumptions

To evaluate the current and future value of the SOMA portfolio, toproject the future interest expense of reserve balances, and to projectthe future interest income from the portfolio, assumptions must bemade about the path of interest rates over the projection period.For this analysis, we rely on the consensus near-term interest rateforecasts from the February 2014 Blue Chip survey combined withthe consensus long-term Blue Chip forecasts from the December2013 Blue Chip survey. The assumed path for the federal funds rateand the yield on the ten-year Treasury note are shown in figure 7.The federal funds rate remains in the 0 to 1/4 percent range throughthe first quarter of 2015, then lifts off during the second quarter.We combine the Blue Chip forecast for the federal funds rate withtheir forecasts for the five-year, ten-year, and thirty-year Treasuriesto construct an entire yield curve.22

3.3 Near-Term Balance Sheet Assumptions

This sub-section reviews our projection methodology for selectedasset and liability items that are of particular interest.

3.3.1 SOMA Portfolio

The evolution of the SOMA portfolio is intended to be consistentwith FOMC communications through April 2014. In particular, weassume the following:

(i) Holdings of securities are increased at a declining pace con-sistent with the February 2014 primary dealer survey, untilpurchases stop in October 2014. The total expansion in hold-ings of Treasury securities and MBS over 2013 and 2014 isabout $1.5 trillion.

22Refer to Carpenter et al. (2013).


Figure 7. Interest Rates

BaselineHigh IRs

Later Liftoff

Federal Funds Rate

2010 2012 2014 2016 2018 2020 2022 2024

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5percent

Quarterly

10-Year Treasury Rate

2010 2012 2014 2016 2018 2020 2022 20241.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

7.0

7.5percent

Quarterly

(ii) Reinvestment of principal payments from agency securitiesinto agency MBS continues until the FOMC begins to unwindthe current accommodative monetary policy stance.


Specifically, maturing or prepaying securities are assumed to be rein-vested until six months prior to the first projected increase in thefederal funds rate.

Given the initial composition of the SOMA portfolio on February28, 2014, the portfolio evolves reflecting these two primary assump-tions and the fact that, over time, securities held in the portfolio age,mature, or prepay. The interest earned on securities already in theportfolio is known. The interest rates on securities purchased in thefuture are based on a projection of all outstanding securities availableto purchase including future issuance. Moreover, the composition offuture purchases imposes the assumed constraint announced by theFederal Reserve Bank of New York that SOMA holdings in any oneCUSIP will remain below 70 percent of the total amount outstandingin that CUSIP.23

In contrast to the Treasury securities held in the SOMA, thematurities of the MBS securities held in the portfolio are a functionof prevailing interest rates given the embedded optionality of themortgages underlying these securities. To capture the effect of inter-est rates on the paydown path of the MBS portfolio, we implementa stripped-down version of the MBS prepayment model proposed inRichard and Roll (1989). We abstract from the seasonality of prepay-ments, since our primary focus is not on the month-to-month varia-tion of prepayments but rather the long-term prepayment behaviorof the portfolio. We also abstract from the “burnout” of prepay-ments; “burnout” refers to the exhaustion of mortgage refinancingsin the MBS, as the loans that remain in the security after a sustainedperiod of refinancing opportunities are unlikely to prepay. Since therates in our set of scenarios are never decreasing, the underlyingmortgages will always be at or moving away from the “money” andwill therefore never experience burnout. The remaining terms of themodel are a seasoning factor and a refinancing incentive factor. Sea-soning captures the observed “ramp” in prepayment behavior. The“ramp” captures the fact that individuals are unlikely to prepaymortgages that were recently issued regardless of the mortgages’“moneyness.” The refinancing incentive factor captures the option-ality component of the underlying mortgages and is defined as a

23Refer to http://www.newyorkfed.org/markets/lttreas faq.html.

http://www.newyorkfed.org/markets/lttreas_faq.html.


function of the ratio of the coupon rate on the MBS to the prevail-ing mortgage rate. The refinance incentive factor takes the followingform:

Refinancing Incentive = .2406 − .1389

∗ arctan(

5.952 ∗(

1.089 − CouponRate

MortgageRate

)).

The parameters of this equation are taken from the Office of ThriftSupervision (2000). Using the combined seasoning and refinancingincentive factors, along with our interest rate paths, we can projectthe cash flows on both current and future MBS holdings, and usethese cash flows to obtain the level and approximate value of theMBS portfolio in any given period.24

It is important to note that Federal Reserve accounting recordsthe securities holdings at face value and records any unamortizedpremium as a separate asset or unamortized discount as a sepa-rate negative asset. Consequently, we project both the face value ofthe portfolio and the associated premiums. To project premiums onsecurities purchased in the future, we calculate the market value ofthe securities at the time of the purchase, which we assume is thepresent discounted cash flow of these securities. To discount the cashflows from Treasury securities, we use the yield curves constructedfrom the Blue Chip forecasts for the federal funds rate, five-yearrate, ten-year rate, and thirty-year rate. To discount the cash flowsfrom the MBS securities, we apply an additional add factor to allpoints along the Treasury yield curve. To calibrate this add factor,we match the realized market value of the MBS held in the port-folio at the end of February to the value of projected cash flowsdiscounted by the Treasury yield curve plus the add factor. For theprojection, we take this add factor and phase it into the long-runhistorical spread of approximately 40 basis points over the next sixmonths.

24Given that this methodology only incorporates a single path of interest ratesand therefore a single path of cash flows for a given scenario, the valuation neglectsthe probability of future “moneyness” of the underlying mortgage and is thereforea simplification of a true MBS valuation.


3.3.2 Liabilities and Capital

In our modeling, projections of Reserve Bank liabilities and capitalare also critical. In the near term, the size of the balance sheet isdriven primarily by securities purchases boosting the asset side ofthe balance sheet and reserve balances increasing on the liabilitiesside as the primary offsetting accounting entry. Later in the projec-tion, normalization of the size of the balance sheet occurs, whichis the point when the liabilities side begins to determine the sizeof the balance sheet. That is, like historical times, reserve balancesbecome fairly small, so increases in currency are the main determi-nant of changes in the size of the balance sheet. For simplicity, weassume that Federal Reserve notes grow in line with the Blue Chipforecast for nominal GDP.25 Capital paid in is assumed to grow atits decade average of 15 percent per year, and surplus is equated tocapital paid in.26 This growth rate plays a role in the long-run trendgrowth rate of the SOMA portfolio.

Until the size of the balance sheet is normalized, we allow reservebalances to be endogenous, calculated as the residual of assets lessother liabilities less capital. When reserve balances fall to the nomi-nal level of $25 billion as the portfolio shrinks, however, we assumethat the Federal Reserve does not allow them to fall further. As cur-rency and Reserve Bank capital are still expanding at that point,purchases of Treasury securities are assumed to restart. Holdings ofTreasury securities expand at the same rate as currency and ReserveBank capital, keeping reserve balances at the assumed $25 billion

25In a classic money demand model with no change in velocity, one can proxymoney growth with nominal GDP growth. That said, there are a number of fac-tors that influence demand for currency beyond nominal GDP, including demandfor currency from abroad, demand for currency during financial crises, and tech-nological change in payment systems.

26In the years prior to the financial crisis, capital paid in grew rapidly. Eachmember bank of the Federal Reserve System is required, by law, to subscribeto shares of its local Reserve Bank in an amount equal to 6 percent of its owncapital and surplus. Of this 6 percent, half is held at the Federal Reserve and theother half is on call at the bank. Consolidation in the banking industry, whichresulted in rapid growth of member bank assets, and regulatory pressures ledto higher key performance indicators from member banks. Member bank assetgrowth declined during the financial crisis; however, capital-paid-in growth mayincrease going forward because of, for instance, systemically important financialinstitution surcharges or Basel III requirements.


level. To maintain reserve balances at $25 billion, we assume that theDesk begins to purchase Treasury bills. Purchases of bills continueuntil these securities comprise one-third of the Federal Reserve’stotal Treasury security holdings—about the average proportion ofTreasury holdings prior to the crisis. Once this proportion of bills isreached, we assume that the Desk buys coupon securities in additionto bills to maintain an approximate composition of the portfolio ofone-third bills and two-thirds coupon securities.

3.4 Exit Strategy Assumptions for the Balance Sheet

We tie our modeling of the normalization of policy to the forecastedinitial increase in the federal funds rate. We rely on the generalprinciples for the exit strategy that the FOMC outlined in the min-utes of the June 2011 FOMC meeting, updated for discussion inthe June 2013 minutes. Specifically, we assume that the reinvest-ment of securities ends six months before the federal funds lifts offfrom the zero lower bound. Although the FOMC guidelines notethat reserve-draining tools will be used prior to raising the fundsrate, to support the implementation of an increase in the federalfunds rate when appropriate, we abstract from this detail in thebaseline projection. The key assumptions used in the baseline andalternative normalization projections are summarized in table 6 inthe appendix.

4. Baseline

With the assumptions in place, this section presents the baselinebalance sheet and income projections, assuming no MBS sales. Thisscenario illustrates one path for monetary policy normalization thatis generally consistent with current FOMC communications. Criti-cal assumptions for this scenario, as well as all other scenarios, arefound in table 6 in the appendix.

4.1 Balance Sheet

Figure 8 presents the projections of key balance sheet line items (thesolid lines). As shown in the top-left panel, SOMA holdings move


Figure 8. Selected Assets and Liabilities of theBalance Sheet

SOMA Holdings

2008 2011 2014 2017 2020 2023

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

5500Billions of dollars

Monthly

Reserve Balances

2008 2011 2014 2017 2020 2023

0

500

1000

1500

2000

2500

3000

3500


Monthly

SOMA Treasury Holdings

2008 2011 2014 2017 2020 2023

0

500

1000

1500

2000

2500


Monthly

SOMA Agency MBS Holdings

2008 2011 2014 2017 2020 2023

0

200

400

600

800

1000

1200

1400

1600

1800

2000


Monthly

BaselineBaseline w/ Sales

Later Liftoff


Table 4. Projected Maturing Treasury Securities,$Billion

2015 $3.52016 $215.52017 $194.22018 $374.0

up slightly through the middle of 2014, reflecting the continuationof the asset purchase program. After the conclusion of purchasesin October 2014, the baseline portfolio begins to decline from itspeak level of $4.2 trillion as securities are allowed to redeem withoutreinvestment. The peak size of the portfolio is much larger than thesize of SOMA immediately prior to the financial crisis, which wasroughly $800 billion, and roughly $3 trillion above Federal Reservenotes.

After purchases end, under the assumption that the FOMCbegins to allow all asset holdings to roll off the portfolio as the firststep in the exit strategy, SOMA holdings begin to decline. How-ever, because the Federal Reserve sold or redeemed almost all of theTreasury securities with less than three years of remaining matu-rity during the maturity extension program in 2011–12, the port-folio holds very few shorter-dated Treasury securities at the timeredemptions begin. Therefore, as shown in the bottom-left panel,when rolloff begins in November 2014, only a minimal amount ofsecurities are maturing, and Treasury securities do not immediatelydecline. As shown in table 4, the amount of Treasury securities thatare maturing becomes sizable in 2016. In particular, between 2016and 2018, nearly $750 billion in Treasury securities are expected tomature and roll off the portfolio.

While Treasury securities do not decline until sometime afterliftoff, MBS holdings, the bottom-right panel, begin to contractimmediately. These holdings decline modestly, as prepayments areprojected to be about $45 billion per quarter around the time ofliftoff, and then slow further as rates rise. By the end of 2025, MBSholdings are roughly $540 billion. Recall that in former ChairmanBernanke’s press conference he noted, “in the longer run, limitedsales could be used to reduce or eliminate residual MBS holdings.”


This projection suggests that residual holdings are still a sizableamount.

The decline in Treasury and MBS securities implies that thesize of the balance sheet is normalized in May 2021 with $1 trillionin Treasury securities holdings and $890 billion in MBS holdings.Afterwards, SOMA begins to expand in line with the growth of cur-rency and capital. Purchases of Treasury securities can be strategicto move the portfolio toward the composition that the FOMC desiresin the longer run.

The level of reserve balances throughout the projection roughlyreflects the asset program minus currency in circulation. As shownin the top-right panel, reserve balances top out at $2.9 trillion inNovember 2014, as the SOMA portfolio peaks with the end of assetpurchases. Further out in the projection, the reduction in the sizeof the SOMA portfolio, along with the projected growth of ReserveBank capital and Federal Reserve notes, results in declines in thelevel of reserve balances. Since we assume that reserve balancesdo not fall below $25 billion, by mid-2021 the Desk again startsto reinvest maturing Treasury securities and begins purchases ofTreasury securities. If one were to consider a higher level of steady-state reserve balances, then normalization would occur slightlyearlier.

4.2 Income

Figure 9 shows the path of Reserve Bank net income. Because ofthe large size of the SOMA portfolio, combined with the (relativelyhigh) coupons on the securities, interest income is elevated for sometime.27 As the SOMA portfolio begins to contract with the assumedsteps in the exit strategy, interest income declines through mid-2021.After reserve balances reach $25 billion, Treasury purchases resume,expanding the portfolio, causing interest income to rise.

Interest expense reflects both the level of the federal funds rateand the level of reserve balances. The federal funds rate in the dealer

27The current weighted average coupon on the SOMA portfolio is 3.4 percent.This weighted average coupon evolves over the projection period as securities arepurchased or are removed from the portfolio.


Figure 9. Income Projections


Later LiftoffIOER +50bp

Interest Income

2010 2013 2016 2019 2022 2025

0

20

40

60

80

100

120


Annual

Interest Expense

2010 2013 2016 2019 2022 2025

0

20

40

60

80

100

120


Annual

Realized Capital Losses

2010 2013 2016 2019 2022 2025

−20

0

20

40

60

80

100

120


Annual

Remittances to Treasury

2010 2013 2016 2019 2022 2025

−20

0

20

40

60

80

100

120


Annual

Deferred Asset

2010 2013 2016 2019 2022 2025

0 10 20 30 40 50 60 70 80 90100110120

Billions of dollars

End of year

Memo: Unrealized Gains/Losses

2010 2013 2016 2019 2022 2025−500

−400

−300

−200

−100

0

100

200

300


End of year

survey begins to rise in 2015, and interest expense rises with it. How-ever, in 2018, interest expense begins to moderate, as the decline inreserve balances more than offsets the rise in the federal funds rate.

On net, annual remittances to the Treasury remain elevated byhistorical standards in the near term, but then decline. The trough


in remittances is $18 billion in 2018, a level that is not much lowerthan the $25 billion average remittances in the decade prior to thefinancial crisis. There is no deferred asset in this baseline projec-tion. Cumulative remittances from 2009 through 2025 are $918 bil-lion, above the level predicted by a trend growth in remittances.Of course, the overall effect on the federal government’s finances ismore complicated than just the impact from Federal Reserve remit-tances. For example, if asset purchases provide meaningful economicstimulus, the increase in government revenues from faster economicgrowth could more than offset any lull in remittances. Further, ifthe asset purchases lower interest rates, the interest expense of thefederal government is lower.

Although only realized gains or losses affect the Federal Reserve’sincome, we project the unrealized gain or loss on the portfolio. Theunrealized loss on the portfolio at a point in time is defined as thedifference between the projected market value of the portfolio, lessthe amortized cost of the portfolio (par value of the securities plusnet premiums). Given the large SOMA portfolio and the projectedrise in interest rates, under the baseline projections, the portfolio isin an unrealized loss position at the end of 2014. This unrealized lossposition continues to grow through the beginning of 2018, but sub-sequently diminishes as the portfolio shrinks through redemptionsand sales.

5. Alternative Normalization Strategies

The baseline assumption of how the FOMC may choose to unwindunconventional monetary policy is one of many alternatives availableto the Committee. Here we consider a few alternative normaliza-tion strategies: MBS sales, alternative forward guidance, and term-draining tools. We compare the effects of these alternative strategieson the balance sheet and income relative to the baseline projection.

5.1 MBS Sales

The June 2011 FOMC minutes laid out exit strategy principlesthat included selling MBS over a period of three to five years atsome date after the funds rate moved above the zero lower bound.


Selling securities is one way of raising interest rates; see, for example,Ihrig et al. (2012). In addition, FOMC members have expressed adesire to remove MBS from the portfolio, in part reflecting their viewthat the Federal Reserve should minimize the extent to which theFederal Reserve portfolio might affect the allocation of credit acrosssectors of the economy.28 In this projection, we consider selling MBSholdings over four years, commencing six months after liftoff. SellingMBS after the funds rate starts to rise is not only a way to removeMBS from the portfolio but also a way to reduce the amount ofunconventional monetary policy in place at a time when the FOMCwants to firm monetary policy. A consequence of selling MBS is thatthe Federal Reserve will realize capital loses, reflecting selling rela-tively low-coupon MBS in an environment with rising interest rates.This type of strategy will reduce remittances to the Treasury.

The implications of MBS sales on the balance sheet are shownin figure 8. With MBS sales (the dashed lines), MBS holdings dropmuch faster than in the baseline. Consequently, the balance sheetwith MBS sales normalizes in size around May 2019, implying thatunconventional monetary policy is unwound one year earlier than inthe baseline.

The income projection is a bit different from the baseline.Because of MBS sales, as shown in figure 9, there are fewer secu-rities in SOMA and so interest income is lower in the medium term.Interest expense is also lower, because of the reduction in reservebalances. Under this path of interest rates, with sales come real-ized capital losses.29 Over the four-year sales period, September2015 to August 2019, these losses average roughly $35 billion peryear. Putting the pieces together, remittances fall to zero from 2017through 2020.

The projection has a deferred asset that peaks at about $45 bil-lion. This illustrates that if policymakers choose to sell MBS, thereis a chance that remittances will be zero for a period of time. More-over, as Greenlaw et al. (2013) have suggested, this could involve

28Refer to the minutes of the April 2011 FOMC meeting, available athttp://www.federalreserve.gov/monetarypolicy/fomcminutes20110427.htm.

29Treasury securities sales conducted under the maturity extension programresulted in small gains because of the low level of market interest rates in 2012and the relatively higher coupon on the securities sold.

http://www.federalreserve.gov/monetarypolicy/fomcminutes20110427.htm.


negative political pressure. However, zero remittances do not meanthe Federal Reserve cannot conduct monetary policy. Other centralbanks have operated with losses. For example, the Swiss NationalBank experienced an operating loss in 2008 and 2010, as a result oftheir currency interventions in support of the Swiss franc.30 Despitethese losses, the ability of the Swiss National Bank to influence mon-etary conditions was relatively unaffected. Moreover, policymakershave suggested this is not their preferred normalization strategy andwould have the choice to slow or stop sales if this policy was foundto be inconsistent with their dual mandate.

Despite the limited period of zero remittances, the level of remit-tances plus capital remains positive for most of that time, suggestingthat the Federal Reserve could see only brief periods of nega-tive equity. In addition, from an operational point of view, zeroremittances do not preclude the FOMC from conducting mone-tary policy. As noted by Cukierman (2011) and others, a centralbank differs from a private corporation in that its objective is notprofit maximization—for example, in the United States, the Fed-eral Reserve’s objective is promoting maximum employment, stableprices, and moderate long-term interest rates—and it can thereforeoperate with negative equity.

For simplicity, this scenario is modeled assuming the same under-lying macroeconomy as in the baseline. Of course, the underlyingeconomy could differ, and most likely would be part of the rea-son that the FOMC chose to deviate from its current plans not tosell MBS in its normalization strategy. For example, it might bethe case that the economy starts to expand at a faster rate thandesired or inflation rises above the Committee’s 2 percent objec-tive. If so, MBS sales could put upward pressure on interest ratesand return the economy to the baseline path. So, assuming that themonetary policy actions are effective, the medium- to longer-run pro-jection for the economy should be similar to what is assumed in thebaseline.

30Refer to “Annual Result of the Swiss NationalBank” for 2008 and 2010, available for download athttp://www.snb.ch/en/mmr/reference/pre 20090304/source/pre 20090304.en.pdfand http://www.snb.ch/en/mmr/reference/pre 20110303/source/pre 20110303.en.pdf.

http://www.snb.ch/en/mmr/reference/pre_20090304/source/pre_20090304.en.pdf

http://www.snb.ch/en/mmr/reference/pre_20110303/source/pre_20110303.en.pdf


5.2 Alternative Forward Guidance Thresholds

From December 2012 to March 2014, the FOMC provided forwardguidance about the federal funds rate in terms of a threshold forthe unemployment rate. The FOMC statement explicitly noted thatthe funds rate would remain in “this exceptionally low range . . .at least as long as the unemployment rate remains above 6-1/2 per-cent.” In the January 2014 FOMC minutes, the Committee contem-plated alternative quantitative forward guidance.31 In March, theminutes noted that a “participant favored introducing new quan-titative thresholds of 51/2 percent for the unemployment rate and21/4 percent for projected inflation.” Here, we consider the impactof implementing a threshold that would push out liftoff from thedate assumed in the baseline.

Lowering the threshold implies shifting market participants’beliefs to a later liftoff of the federal funds rate. For illustrativepurposes, we assume the liftoff occurs four quarters later than thebaseline, which is when the Blue Chip forecast has the unemploy-ment rate reach 5.6 percent and CPI inflation one year out is 2.3percent. This later liftoff implies that the contour of the balancesheet will change, delaying the decline in the portfolio and thereforethe normalization of the size of the balance sheet. In addition, thedelay in liftoff affects income. Of course, a critical question is howfast rates will rise after liftoff. We assume the funds rate moves upat the same pace as the baseline scenario, as illustrated in figure 7.We also assume that rolloff begins six months before liftoff, delayingthe start to rolloff by six quarters from the baseline. The ten-yearyield is adjusted by a simple approximation of the expected changein the rate as implied by the expectations hypothesis. That is, welower the ten-year yield by the average decrease in the path of thefederal funds rate over the next forty quarters.

Figure 8 illustrates the evolution of the balance sheet (thedashed-dotted lines). The delayed start to stopping reinvestmentimplies larger MBS holdings throughout the projection period, withabout $600 billion in holdings at end-2025.32 For Treasury securities,however, the delayed start to allowing the securities to roll off the

31 FOMC minutes are found here: http://www.federalreserve.gov/monetarypolicy/fomccalend32Chairman Bernanke mentioned in his press conference statement that resid-

ual agency MBS holdings could be sold at some point in the future.

http://www.federalreserve.gov/monetarypolicy/fomccalendars.htm.


portfolio when they mature is not as dramatic. This is in part a resultof the maturity extension program, in which the Federal Reserve soldor allowed to redeem all securities with remaining maturity of lessthan three years and purchased the same amount of securities withremaining maturity of six years or greater over the course of 2011and 2013. Consequently, there are few securities maturing in 2015(see table 4). Hence, projected Treasury holdings in the mediumterm are not that different from the baseline. Of course, later in theperiod, Treasury holdings are less than the baseline since there aremore MBS holdings in this scenario. Taken together, the evolutionof the securities holdings implies that normalization of the size ofthe balance sheet is delayed by six months relative to the baseline,implying a longer period for unconventional monetary policy to bein place.

Figure 9 shows that this policy would boost remittances to theTreasury by a sizable amount. Interest income is boosted throughthe medium run by the higher securities holdings. Interest expenseis generally lower than the baseline, reflecting the fact that delayedstart to the rise in the federal funds rate allows more Treasury secu-rities to roll off the books and reduce reserve balances faster oncethe federal funds rate rises. These two factors imply that remittancesare much higher through the medium term, with a trough of roughly$25 billion. Cumulative remittances are $989 billion, $71 billion morethan the baseline. This scenario shows that if the FOMC chose tolower the threshold, for whatever reason, unconventional monetarypolicy would be unwound a bit more slowly, while remittances wouldbe boosted relative to the baseline scenario. Again, MBS holdingswould be sizable at the end of 2025.

Of note, our analysis abstracts from possible macroeconomiceffects of delaying the rise in rates. For example, while there could bebeneficial effects on output if rates were held “low for long,” inflationcould rise substantially. In particular, evidence from DSGE modelssuggests a particularly outsized response from forward guidance, andsome authors suggest methodologies for damping this response (DelNegro, Giannoni, and Patterson 2013). Moreover, there could be evi-dence of excess risk taking in financial markets, leading to financialinstabilities, discussed in Feroli et al. (2014). The conditionality offorward guidance should mitigate these risks; however, as with any


policy decision, there exists some uncertainty and therefore somerisk of an unfavorable outcome.

5.3 Term Reserve-Draining Tools

So far, our analysis has assumed that the Federal Reserve has notengaged in any active liability management and, as a result, reservebalances passively decline as securities mature and roll off the port-folio. As noted in the June 2011 exit principles, the Committee mayelect to incorporate liability management tools to reduce or “drain”reserve balances into its exit strategy in order to support condi-tions in which the federal funds rate trades near the intended targetpolicy rate. Tools that could be used to drain reserve balancesinclude reverse repurchase agreements and term deposits. RRPscan be conducted at an overnight or term frequency. OvernightRRPs would result in the balance sheet composition shifting fromreserve balances to RRPs, but with the overnight rate likely beingnear IOER, there would likely be only a minimal effect on income.For term-draining tools, however, the income effect could be morenoticeable.

If the Federal Reserve were to use term-draining operations wherecounterparties demanded a relatively high rate of return since theseterm operations have a longer maturity and would be less liquidthan reserve balances, interest expense would rise. To illustrate thispoint, we assume that all reserve balances pay 50 basis points aboveIOER. This scenario is calibrated to one standard deviation of thehistorical spread between the federal funds rate and selected one-and three-month money-market rates.

The size of the balance sheet is unchanged in this scenario,though there would be a shift in the composition of liabilities: reservebalances would fall 1:1 with the use of term deposits and term RRPs.Interest expense would rise, with an increase of 50 basis pointsper each dollar drained. Given we assume all reserve balances aredrained, an extreme example, as shown in table 5 and figure 9, evenin this case, annual remittances are only marginally affected. Thisresult is because the balance sheet is shrinking at the time inter-est expense is rising. The impact of higher costs reduces cumulativeremittances by about $40 billion. Given the magnitude of the other


Table 5. Projected Remittances, $Billion

Cumulative2015 2016 2017 2018 2019 2020 2009–2025

Baseline 93.8 53.3 25.2 18.2 22.9 31.0 918.4Costly 83.8 41.7 15.8 11.4 18.9 29.3 875.1

Draining*

*Term-draining tools implemented on all reserve balances from liftoff to when reservebalances are normalized.

costs and revenues, the expense associated with draining tools doesnot seem too large.

6. Interest Rate Sensitivity—Deferred Asset

Above we found that with baseline assumptions the Federal Reservewill normalize monetary policy and its balance sheet without anydeferred asset. The only way that remittances could fall to zero inthis case would be if interest expense rose sustainably while reservebalances were elevated. In the MBS sales scenario, remittances couldbe halted for a longer period than projected above if realized losseswere larger. To illustrate these points, we allow interest rates torise 200 basis points higher after liftoff than in the baseline projec-tion. The results will highlight the point discussed in the December2012 minutes: “Depending on the path for the balance sheet andinterest rates, the Federal Reserve’s net income and its remittancesto the Treasury could be significantly affected during the period ofpolicy normalization.” In particular, it could be necessary to raiserates faster than in the baseline if the economy was overheating orif inflation was consistently well above the Committee’s 2 percentobjective. This shock to interest rates has two effects on the size ofthe balance sheet. First, higher interest rates reduce the incentive formortgage holders to refinance, causing MBS prepayments to slow.Second, higher interest rates increase interest expense on reserve bal-ances in both scenarios and realized losses in the scenario with sales.All these factors increase the likelihood of a deferred asset, whichalso delays the date of normalization.


Figure 7 shows the projected rates for the higher interest ratescenarios. The federal funds rate and ten-year Treasury yield riseat a faster pace at liftoff, and after one year are 200 basis pointshigher than the baseline rates over the remainder of the projection.In the baseline interest rate projection, the ten-year Treasury yieldrises by approximately 1 percentage point between end-2014 andend-2016. By contrast, the 200-basis-point shock implies that theten-year Treasury yield is increasing by 3 percentage points overthose two years.

There are a couple of ways to put the size of this shock in per-spective. To start, this size shock is 1.3 percentage points above theaverage forecast of the top 10 highest respondents in the Decem-ber 2013 Blue Chip survey (roughly 20 percent of the sample), andthus is probably comfortably above most market participants’ inter-est rate projections. In addition, for a historical comparison, from1978 to present, the standard deviation of the two-year change inthe ten-year Treasury yield is 1.6 percentage points. As a result, thishigher interest rate scenario should be seen as a somewhat unlikelyscenario, but not an implausible one. Of course, to the extent thatinflation expectations have become better anchored through time,this increase in interest rates may be even less probable than thehistorical record may suggest.

Focusing on the baseline, no-MBS-sales scenario, shown in figure10, the interest rate shock does not substantially change the Fed-eral Reserve’s balance sheet projections.33 The income projection,as shown in figure 11, does change, however. The higher federalfunds rate implies greater interest expense. Once combined withnon-interest income and expenses, remittances to the Treasury fallto zero for several years and a deferred asset is booked for 2017through 2020.

How should we interpret this shock? Cumulative remittancesfrom 2009 to 2025 are $791 billion, about $125 billion less thanin the baseline. To put some perspective on this value, it remainsgreater than what would be suggested by a pre-crisis 1990–2007trend level of remittances. This projection is also similar towhat Christensen, Lopez, and Rudebusch (2013) report as their

33A deferred asset will have a small impact on the size of the SOMA portfolio,but not enough to see in the figures.


Figure 10. Selected Assets and Liabilities ofthe Balance Sheet

SOMA Holdings

2008 2011 2014 2017 2020 2023

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000


Monthly

Reserve Balances

2008 2011 2014 2017 2020 2023

0

500

1000

1500

2000

2500

3000

3500


Monthly

SOMA Treasury Holdings

2008 2011 2014 2017 2020 2023

0

500

1000

1500

2000

2500


Monthly

SOMA Agency MBS Holdings

2008 2011 2014 2017 2020 2023

0

200

400

600

800

1000

1200

1400

1600

1800

2000


Monthly


High IRsHigh IRs w/ Sales


Figure 11. Income Projections


High IRsHigh IRs w/ Sales

Interest Income

2010 2013 2016 2019 2022 2025

0

20

40

60

80

100

120


Annual

Interest Expense

2010 2013 2016 2019 2022 2025

0

20

40

60

80

100

120


Annual

Realized Capital Losses

2010 2013 2016 2019 2022 2025

−20

0

20

40

60

80

100

120


Annual

Remittances to Treasury

2010 2013 2016 2019 2022 2025

−20

0

20

40

60

80

100

120


Annual

Deferred Asset

2010 2013 2016 2019 2022 2025 0

50

100

150

200

250


End of year

Memo: Unrealized Gains/Losses

2010 2013 2016 2019 2022 2025−700−600−500−400−300−200−100 0 100 200 300 400

Billions of dollars

End of year

lower-bound “probability-based” stress scenario.34 Further, theynote that the chance that cumulative remittances would be belowan underlying trend value is less than 0.1 percent.

34 Christensen, Lopez, and Rudebusch (2013) suggest that the probability ofthis shock is less than 5 percent.


Turning to a scenario where MBS are sold, the higher interestrate path does not change the balance sheet by much, but withhigher interest expense and larger capital losses, a deferred assetpeaks at nearly $263 billion. Moreover, remittances to the Treas-ury are halted for 83/4 years. Cumulative remittances from 2009to 2025 are $635 billion, about $283 billion less than in the base-line. Of course, the June 2013 minutes stated that policymakersdo not expect to sell agency mortgage-backed securities during theprocess of normalizing monetary policy. Therefore, this scenario isvery unlikely to play out. In addition, policymakers could changetheir exit strategy, slowing or stopping these asset sales.

These sensitivity scenarios illustrate that in some circumstancesthe Federal Reserve could have years with no remittances to theTreasury and a deferred asset on its books. It is important, however,that these scenarios be viewed within a macroeconomic framework.As noted above, to the extent that asset purchases are effectivein stimulating the economy, overall government revenues would beboosted on net, despite the capital losses at the Federal Reserve. Inaddition, one should consider the Federal Reserve’s remittances overthe entire period of unconventional monetary policy. Overall, aver-age annual remittances to the Treasury even in these shock scenar-ios remain well above the average annual remittances of $25 billionrecorded prior to the crisis.

More broadly, as mentioned above, zero remittances would notaffect the conduct or stance of monetary policy. This assertion isbased on a number of points that have been raised previously in theacademic literature. First, monetary policy in the United States hashistorically been conducted by adjusting short-term interest rates,and there is no direct, causal link between the remittances to theTreasury and short-term interest rates.

Second, Cukierman (2011) and others cite that the main driversof central bank insolvency include assumptions of problem assets bythe central bank, fiscal abuse, and a buildup of large stocks of foreignreserves, sometimes to support an exchange rate peg. These driversare absent from the U.S. case. Other drivers that could be moreapplicable to the United States are a need to absorb banking-sectorexcess liquidity and central bank purchases of relatively low-yieldinginstruments. That said, in most scenarios, we find that any deferred


asset would most likely be for a handful of years and would notpresent an ongoing concern.

And third, one possible (though in our view, unlikely) channelthrough which losses could impair monetary policy is if, for somereason, economic agents believed that the central bank’s earningsaffected inflation. If beliefs were formed in that way, perhaps becauseof a misunderstanding of the mechanics of the economy, inflationexpectations could rise and thereby become embedded in actualinflation. Relatedly, a model by Hall and Reis (2013) provides atheoretical model that explores the potential for a central bank topursue an inflationary policy in order to erode capital losses. Theystate that the realization of this outcome in the United States isunlikely and find the potential for the Federal Reserve to becomeinsolvent to be “remote.” By contrast, empirical evidence providedby Klueh and Stella (2008) shows that in some countries, there canbe a negative relationship between inflation and central bank cap-ital. Their results were based on a set of countries where centralbank losses as a percent of nominal GDP approached 2 percent, wellbelow any levels projected here.

Such a process of a feedback loop of inflation and losses seemsunlikely for the case of the United States. In particular, authorssuch as Del Negro and Sims (2014) point to the importance of thepresent value of net worth, rather than of equity, as the key measureof central bank solvency, and perhaps by extension, central bankindependence. Because the Federal Reserve has an expected (posi-tive) stream of seigniorage revenue from its currency franchise, thepotential impact of a short-lived deferred asset on inflation expec-tations or on the necessity of the central bank to be recapitalizedby the Treasury should be non-existent, or negligible. That said,Del Negro and Sims do highlight extraordinary circumstances ofalternative equilibria under which the presented discounted value ofseigniorage revenue is not sufficient to cover the gap between theinterest earned on assets less the interest paid on liabilities. Thiscould be a result of inflation expectations becoming embedded inasset values, thereby eroding their present worth. In their model, acentral bank would have to resort to fulfilling those expectations byallowing inflation to drift above its target. However, in order for thisto occur, the balance sheet would likely have to be multiples of thesize of the current balance sheet, an unlikely outcome.


Moreover, given this apparatus to produce income projectionsfor any monetary policy scenario, policymakers could evaluate alter-native ways to remove monetary policy accommodation with theidea of choosing the path that most likely reduces the possibility ofzero remittances and, hence, political scrutiny of their actions. Twoadditional strategies not presented here, for example, but that couldreduce the probability of zero remittances are security sales in thenear term when remittances are robust and committing to a moregradual pace of tightening after liftoff. Looking across all possiblestrategies, there would be trade-offs between alternative monetarypolicy actions, their effect on economic activity, and remittances.When considering these alternative actions, the FOMC would needto keep in mind its dual mandate of full employment and pricestability.

7. Conclusion

In this paper, we have outlined a variety of ways the FOMC mayunwind the unconventional monetary policy that it has institutedover the past several years. The different policies have implicationsfor the length of time unconventional policy is in place, the compo-sition of the Federal Reserve’s balance sheet, and remittances to theTreasury. How fast unconventional monetary policy unwinds and thetools used depends on FOMC actions, and we discussed the impactof a few of these possibilities.


Appendix

Table 6. Key Assumptions of the Projections

Baseline LaterAssumption Baseline with Sales Liftoff

Current Portfolio Strategy

Agency Reinvestments Agency MBS Agency MBS Agency MBS

Treasury Purchases

Total Amount (2013–14) $790 billion $790 billion $790 billionJan. 2013 to Dec. 2013 45 45 45Jan. 2014 40 40 40Feb. 2014 to Mar. 2014 35 35 35Apr. 2014 30 30 30May 2014 to Jun. 2014 25 25 25Jul. 2014 20 20 20Aug. 2014 to Sep. 2014 15 15 15Oct. 2014 10 10 10

MBS Purchases

Total Amount (2013–14) $680 billion $680 billion $680 billionJan. 2013 to Dec. 2013 40 40 40Jan. 2014 35 35 35Feb. 2014 to Mar. 2014 30 30 30Apr. 2014 25 25 25May 2014 to Jun. 2014 20 20 20Jul. 2014 15 15 15Aug. 2014 to Sep. 2014 10 10 10Oct. 2014 5 5 5

Exit Strategy

Fed. Funds Liftoff 2015:Q2 2015:Q2 2016:Q2Redemptions Start Nov. 2014 Nov. 2014 Nov. 2015Sales Start N/A Oct. 2015 N/ASales End N/A Sep. 2019 N/A


References

Carpenter, S., J. Ihrig, E. Klee, A. Boote, and D. Quinn. 2013.“The Federal Reserve’s Balance Sheet and Earnings: A Primerand Projections.” FEDS Paper No. 2013-1 (January), Board ofGovernors of the Federal Reserve System.

Christensen, J., J. Lopez, and G. Rudebusch. 2013. “A Probability-Based Stress Test of Federal Reserve Assets and Income.” Work-ing Paper No. 2013-38 (December), Federal Reserve Bank of SanFrancisco.

Cukierman, A. 2011. “Central Bank Independence and MonetaryPolicymaking Institutions — Past, Present and Future.” Euro-pean Journal of Political Economy 24 (4): 722–36.

Del Negro, M., M. Giannoni, and C. Patterson. 2013. “The ForwardGuidance Puzzle.” Staff Report No. 574 (May), Federal ReserveBank of New York.

Del Negro, M., and C. Sims. 2014. “When Does a Central BankRequire Fiscal Support?” Mimeo (March).

Edwards, C. E. 1997. “Open Market Operations in the 1990s.” Fed-eral Reserve Bulletin (November): 859–74.

Federal Reserve Bank of New York. 2013. “Domes-tic Open Market Operations in 2012.” Available athttp://www.ny.frb.org/markets/omo/omo2012.pdf.

Feroli, M., A. Kashyap, K. Schoenholtz, and H. S. Shin. 2014. “Mar-ket Tantrums and Monetary Policy.” Chicago Booth WorkingPaper No. 14-09 (February).

Friewald, N., R. Jankowitsch, and M. G. Subrahmanyam. 2014. “ToDisclose or Not to Disclose: Transparency and Liquidity in theStructured Product Market.” Mimeo, NYU Stern School of Busi-ness (February 15).

Greenlaw, D., J. D. Hamilton, P. Hooper, and F. Mishkin. 2013.“Crunch Time: Fiscal Crises and the Role of Monetary Policy.”(February 22).

Hall, R. E., and R. Reis. 2013. “Maintaining Central-Bank Solvencyunder New-Style Central Banking.” Mimeo.

Ihrig, J., E. Klee, C. Li, B. Schulte, and M. Wei. 2012. “Expecta-tions about the Federal Reserve’s Balance Sheet and the TermStructure of Interest Rates.” FEDS Paper No. 2012-57, Board ofGovernors of the Federal Reserve System.

http://www.ny.frb.org/markets/omo/omo2012.pdf.


Judson, R., and R. Porter. 1996. “The Location of U.S. Currency:How Much Is Abroad?” Federal Reserve Bulletin 82 (October):883–903.

Klueh, U. H., and P. Stella. 2008. “Central Bank Financial Strengthand Policy Performance: An Econometric Evaluation.” IMFWorking Paper No. 08/176.

Krishnamurthy, A., and A. Vissing-Jorgensen. 2011. “The Effectsof Quantitative Easing on Interest Rates: Channels and Impli-cations for Monetary Policy.” Brookings Papers on EconomicActivity (Fall).

Li, C., and M. Wei. 2013. “Term Structure Modeling with SupplyFactors and the Federal Reserve’s Large-Scale Asset PurchasePrograms.” International Journal of Central Banking 9 (1): 3–39.

Meltzer, A. 2010. A History of the Federal Reserve, Vol. 2, Book 1,1951-1986. Chicago: University of Chicago Press.

Office of Thrift Supervision. 2000. ”Net Portfolio Value Model Man-ual.”

Rudebusch, G. D. 2011. “The Fed’s Interest Rate Risk.” EconomicLetter (Federal Reserve Bank of San Francisco) 2011-11 (April11).

Richard, S. F., and R. Roll. 1989. “Prepayments on Fixed-RateMortgage-Backed Securities.” Journal of Portfolio Management15 (3): 73–82.

Vickrey, J., and J. Wright. 2013. “TBA Trading and Liquidity inthe Agency MBS Market.” Economic Policy Review (FederalReserve Bank of New York) 19 (1).

the federal reserve's balance sheet and earnings: a · pdf filevol. 11 no. 2 the federal...

Documents